Chapter content · Overview diagram
Ch 2 — System Four Is Not a Pair of Eyes
Part I — Inside the model. The manuscript charges that System Four, taught as an outward-looking scanner, is a reduction the community reproduced rather than inherited, and that it is not the function Beer wrote about.
The question
Is System Four, as we draw it — a pair of eyes that scans the environment — the function Beer specified?
The finding
The chapter assembles a teaching sample (25 independent web texts, four practitioner textbooks, eight pages of the community’s own training material) and reports that thirteen of the 25 web texts and all eight training pages describe System Four chiefly as scanning or looking outward, while only three say it holds a model of the organization itself. The manuscript calls this a convenience sample from one search engine: it describes what we have taught and found taught, not the field. It scopes the claim to the teaching — what is named, not what the system does — and notes that “scanning” is not Beer’s word: his names for the function are the “outside-and-then” and the “development directorate”, while the scanning vocabulary is borrowed from environmental scanning in strategic management and from horizon scanning in futures studies.
Read against Beer, System Four is larger — the modelling and anticipation capacity that holds an account of the environment and the system’s own relation to it, forms expectations and revises them fast enough to be usable. The chapter reads the loop between System 3 and System 4 as carrying predictions rather than reports alone: what System Four sends into the homeostat is a prediction about what the operations will show next, and an instruction issued on its strength is a wager framed so that the next round of operational reporting can settle it. It marks that reading as its own, supported by Beer’s passages, and records that the phrase “prediction channel” occurs in none of the eight books searched. It does not claim a channel from System Four to the command channel: Beer routes System Four’s indications into System Five, and the adaptation link he names is the Three–Four homeostat.
It names a load-bearing substrate — a reticular-formation model associated with Kilmer, Blum and McCulloch (1969), which Beer lists in his bibliography — and says the report’s actual claims remain to be verified. It declines the argument that old anatomy invalidates the organizational logic. A textual search of eight Beer books returned a negative result: no function distinct from System Four appears to perform forward modelling, and the terms “forward model” and “prediction channel” occur in none of them. The manuscript marks this provisional, since OCR can hide hits and the 1972 edition was not searched. A claim that action-selection logic has run for some 560 million years is recorded as under verification, and the chapter says the number comes out if it does not survive.
What the chapter puts at risk
- System Four must sense the environment.
- System Four must maintain a model.
- System Four must anticipate.
- The System 3–System 4 homeostat carries predictions, not only reports.
- The five-system partition is complete (where does action selection reside?).
- Neural organization licenses organizational design.
Each is put as an open question about what survives if the 1969 substrate is replaced by the modern account. The chapter does not answer them; answering them, it says, is the work it asks for.
Beer locators
BOTF 1981: 181, 182–83, 183, 189, 196; HOE 1979: 229, 235–36, 237, 240; DTS 1985: 111, 115–16, 120; BOTF 1981 Select Bibliography: W. L. Kilmer, J. Blum and W. S. McCulloch, The Reticular Formation (1969).
Status
Draft, Punch and Judy v0.5.0 — not project canon, and not published. The search result is a provisional negative (OCR full text of eight Beer books, 24 Sept 2026; the 1972 edition unsearched). The Kilmer/Blum/McCulloch report’s claims are unverified. The 560-million-year claim is marked under verification. The teaching sample is a convenience sample, not a measure of the field. Section 1 and section 2 were revised, and a footnote added on the provenance of “scanning”, in the comment review of 12 Oct 2026.
Related
System Four, Viable System Model, Variety and regulation, Stafford Beer, Punch and Judy — Overview, Ch 1 — The Mapping We Teach, Ch 3 — The Law We Quote, Ch 9 — Why None of This Ever Cost Us Anything